Evaluation of a diagnostic algorithm for sputum smear–negative pulmonary tuberculosis in HIV-infected adults

Background: The Revised National TB Control Program bases diagnosis of tuberculosis (TB) on sputum smear examination and response to a course of antibiotics, whereas World Health Organization recommends early chest radiography [chest x-ray (CXR)] for HIV-infected symptomatic patients. We evaluated the utility of initial CXR in the diagnostic algorithm for symptomatic HIV-infected patients with negative sputum smears. Methods: HIV-infected ambulatory patients with cough or fever of ≥2 weeks and 3 sputum smears negative for acid-fast bacilli were enrolled in Chennai and Pune, India, between 2007 and 2009. After a CXR and 2 sputum cultures, a course of broad-spectrum antibiotics was given and patients were reviewed after 14 days. Sensitivity, specificity, positive and negative predictive values of symptoms, CXR, and various combinations for diagnosing pulmonary tuberculosis (PTB) were determined, using sputum culture as gold standard. Results: Five hundred four patients (330 males; mean age: 35 years; median CD4: 175 cells per cubic millimeter) were enrolled. CXR had a sensitivity and specificity of 72% and 57%, respectively, with positive predictive value (PPV) of 21% and negative predictive value (NPV) of 93% to diagnose PTB. TB culture was positive in 49 of 235 patients (21%) with an abnormal initial CXR and 19 of 269 patients (7%) with a normal CXR (P < 0.001). Sensitivity and specificity of cough ≥2 weeks for predicting PTB was 97% and 6%, with PPV and NPV of 14% and 94%, respectively. Conclusions: Although moderately sensitive, basing a diagnosis of TB on initial CXR leads to overdiagnosis. An absence of weight loss had a high NPV, whereas none of the combinations had a good PPV. A rapid and accurate diagnostic test is required for HIV-infected chest symptomatic